Web doublers against bolt bearing & IdeaStatica problem

[MburakY]

Greetings,

I need some tips on solving a specific problem that keeps bugging my mind. We are working as a detailer in a rather large international project. My problem is with how to strengthen a beam web against bolt bearing?

So in this particular project, there are some beams connected to huge girders with wide flanges. These connected beams, have both very high shear and axial forces. Don’t ask me why, that’s what designer provided. And funnily enough these connection forces are nearing almost the beams respective capacity. Anyway, since we have both high shear and axial forces, we used shear splices to connect them to girders. (see attachment1).

Idea was simple, a plate will protrude from girder web, welded to both flanges of the girder as well as its web. Beam end will be placed against it and web splice plates will connect gusset to beam web from both sides. That way we will not have any in plane bending due to axial force as in the case with simple shear tabs. Also we would benefit from double shear effect. Anyway, problem is we often require web strengthening because of bolt bearing.

I also need to mention that owner refuses to use site welding or having flanges connected since it alters structural behavior. They have their reasons but we argued to find a solution but to no avail.

Anyway, my question is that would having web doublers work against this problem? What I would normally do is to find bolt forces due to axial+shear+moment (from eccentricity), check it against the thickness of web + doubler plates and be done with it. Depending on the how much thickness I require, I would use that much of doubler plates. However, we recently purchased a connection software, IdeaStatica to see what it does and find out if we could use. We normally do all calculations by spreadsheets by the way. For a specific example, we have 12 mm we. What we get from the software is that thinner doubler plates somehow takes higher bearing force. From strain compatibility,thicker plates should get higher force but it seems program takes bolt bending into account so doublers strained more than the web, hence higher stresses. I know it all depends on the FEM approach, how stuff is modeled there but it still bugs my mind. I could not find a research on the subject. Do you think this approach is realistic? I want to believe that the old way should work but I need some opinions on that. I attached some more pictures to show what I am trying to say. Also I put the calc. file for those who are interested in the subject.

I hope I made myself clear enough. Please do not hesitate to ask further questions. I would be very happy if someone points me to a research on "force transfer from bolts to multiple plates with varying thicknesses" or something similar. (I think I am terrible at finding keywords.)

 

[KootK]

I want to believe that the old way should work but I need some opinions on that.

My opinion is that the old way should work. As with most things in steel connections, we're assuming a plastic, ductile resisting mechanism that satisfies equilibrium. If either the doubler plates or the beam web need to yield out of the way for the assumed load distribution to manifest itself, so be it. It's not as though all of the holes for the various elements are likely to line up perfectly anyhow.

I'm simultaneously impressed and annoyed that your software examines the elastic state in such detail. Very recently there was a thread here inquiring if FEM would ever be able to assume 100% design of steel connections. Clearly, we still have a ways to go. Or maybe we don't if we're willing to accept the "new way" as a replacement for the old, effectively sacrificing material efficiency for design generality.

 

[MburakY]

Thank you for the reply. I should mention that software does elastoplastic analysis and claims to capture stress redistributions correctly. It also does second orders analysis to capture geometric nonlinear behavior. This is another concern because it is practically impossible to hand check what it does.

 

[rb1957]

I think it odd that the doubler should be higher stressed than the web. In my experience the flexibility of the fasteners prevents the doubler working at the same stress as the web.

"And funnily enough these connection forces are nearing almost the beams respective capacity." they would have called this "efficient design" ... working to the capacity of the beam. However it makes the connections much harder and less likely to be "pinned". But then they could be counting on fixed ends. If the beam is carrying it's shear capacity it becomes "tricky" to extract that load into a connection without overloading.

Are there cap splices as well ? (to create the necessary bending loadpath)


another day in paradise, or is paradise one day closer ?

 

[MburakY]

Cap splices would be the best way to go but as I said, somehow, we get high tensile forces accompanying those high shear forces. Combined with moment due to eccentricity, very thick end plates are usually required alongside with haunch and lots and lots of bolts. I believe forces from designer are exaggerated. They give us a single force for all similar beams where maximum axial and shear coincide, which is rarely the case in reality. Additionally, this is a non orthogonal structure - stadium, meaning that because of the angles between beams, end plates are very difficult to apply.

About doublers and force distributions, I wrote an e mail to the support staff expecting their take on the issue.

 

[CANPRO]

MburakY - I suspect you have concentrated bearing on the double plates because of bolt bending. If the bolt bends slightly you'll lose contact with the beam web on the concave side of the bolt and bearing will be concentrated on the outside edges of the doubler plates. As mentioned by others, this kind of behavior should be expected on some level as load is applied and the bolts kind of "settle in" to where they want to be. I've trialed that program and it seems to be almost too accurate for its own good - seems like it could save you a bit of material here and there, but then you have to deal with "failures" that aren't considered by traditional methods.

Regarding the forces - the axial loads are likely due to lateral load transfer. Some EOR's will give you the actual load in the beams - with the axial load increasing as you move closer to your braced bays - some will give you the max load for one beam and specify it everywhere...sounds like you got the latter. I'm assuming you were given factored loads for design, if so, this is how I deal with the problem you're facing - the shear criteria (whether given as 1/2 Vr or shear based on full load on beam, etc...) is rooted in factored gravity loads (say 1.25DL + 1.5LL) where the axial loads are likely the result of either wind or seismic. If you take both factored loads and apply them to your connection simultaneously, your design is over-conservative (in my opinion). You're basically designing for 1.25DL + 1.5LL + 1.4WL or something similar depending on your code. Typically I'll start with a reasonable connection that I feel should work, I'll do a quick and dirty check based on full factored axial/shear - if that returns a ridiculous connection I'll spend some time coming up with a reasonable and conservative way to "unfactor" and re-factor the axial + shear to appropriate load combinations.

 

[bhiggins]

You need to check the numerical outputs of the software you are using. The bearing capacities of the bolts with doubler plates are significantly lower than the bolts without:

It appears that the program is calculating bearing capacity against the thinnest material (1/4" thickness).

Another common sense check is that all of the bolts connecting to the column are passing. You have more bolts connected to the beam side yet these bolts are failing. You shouldn't require more bolts on one side of the connection vs the other, the design seems inefficient.

Conventional connection design would assume the bearing stress of all the bolts is equal. You should be able to design this connection using hand calcs.

 

[CANPRO]

You shouldn't require more bolts on one side of the connection vs the other, the design seems inefficient.

I disagree. It may be the case that often the same number of bolts will be sufficient on both sides of the connection, but that isn't automatic. It may be a shear connection assumed to transfer no moment, but when you look at the design of the connecting plates, one side of the connection (or both) have to deal with the eccentricity. Which means the stiffer side has to deal with rotation in the bolt group.

With that said, you can see that in the results posted. The bolts that are failing are in the larger bolt group, at the outer edges where they would take the brunt of the rotation. Results do a look a little odd though - why is bolt B24 doing so little compared to other bolts located symmetrically? To resolve the issue at hand, maybe increasing the bolt pitch slightly (if you have room) could help reduce the additional shear from the torsion.

 

[MburakY]

Bolt forces are consistent with our manual checks. There are deviations due to crudeness of our hand checks (ic or elastic method) but for me, I was satisfied. Bolts at the column side are less because they are subjected to less eccentricity.

I consulted with the helpdesk. They have agreed that it is a bug that bearing checked against the thinnest plate. I further consulted them about the force distribution between web and doubler plates and asked for validation etc. I will post if they respond.

We sorted this bug issue with client and they will accept our solution. Basically, although this is very labor intensive, we will replace the part of the web with thicker plate. Like instead of having 6+12+6 we will replace the part of the web with a single 24 mm plate (just in FEM model,not in reality). I believe idea statica will resolve bearing issue in next versions although they said it might take weeks.